Clinical Evaluation of a Fully-automatic Segmentation Method for Longitudinal Brain Tumor Volumetry

Raphael Meier, Urspeter Knecht, Tina Loosli, Stefan Bauer, Johannes Slotboom, Roland Wiest, Mauricio Reyes, Raphael Meier, Urspeter Knecht, Tina Loosli, Stefan Bauer, Johannes Slotboom, Roland Wiest, Mauricio Reyes

Abstract

Information about the size of a tumor and its temporal evolution is needed for diagnosis as well as treatment of brain tumor patients. The aim of the study was to investigate the potential of a fully-automatic segmentation method, called BraTumIA, for longitudinal brain tumor volumetry by comparing the automatically estimated volumes with ground truth data acquired via manual segmentation. Longitudinal Magnetic Resonance (MR) Imaging data of 14 patients with newly diagnosed glioblastoma encompassing 64 MR acquisitions, ranging from preoperative up to 12 month follow-up images, was analysed. Manual segmentation was performed by two human raters. Strong correlations (R = 0.83-0.96, p < 0.001) were observed between volumetric estimates of BraTumIA and of each of the human raters for the contrast-enhancing (CET) and non-enhancing T2-hyperintense tumor compartments (NCE-T2). A quantitative analysis of the inter-rater disagreement showed that the disagreement between BraTumIA and each of the human raters was comparable to the disagreement between the human raters. In summary, BraTumIA generated volumetric trend curves of contrast-enhancing and non-enhancing T2-hyperintense tumor compartments comparable to estimates of human raters. These findings suggest the potential of automated longitudinal tumor segmentation to substitute manual volumetric follow-up of contrast-enhancing and non-enhancing T2-hyperintense tumor compartments.

Figures

Figure 1. Segmentation of necrosis (red), edema…
Figure 1. Segmentation of necrosis (red), edema (green), enhancing tumor (yellow) and non-enhancing tumor (blue) in patient “d”.
From top to bottom, consecutive time points are shown. From left to right: T1 post-contrast-weighted image, FLAIR-weighted image overlayed with segmentation generated by BraTumIA, segmentation generated by Rater-1 and segmentation generated by Rater-2.
Figure 2. Volumes ( n = 64)…
Figure 2. Volumes (n = 64) and volume differences (n = 50) as measured by BraTumIA are plotted against estimates of Rater-1 (circles, blue line) and of Rater-2 (crosses, violet line).
Figure 3. Volumetric evolution of non-enhancing T…
Figure 3. Volumetric evolution of non-enhancing T2-hyperintense tissue (NCE-T2, solid line) and contrast-enhancing tumor (CET, dashed line) over time for BraTumIA (red), Rater-1 (blue) and Rater-2 (violet) shown for one patient.
Figure 4. Volumetric evolution of relative values…
Figure 4. Volumetric evolution of relative values with respect to preoperative volume of non-enhancing T2-hyperintense tissue (NCE-T2, solid line) and contrast-enhancing tumor (CET, dashed) over time for BraTumIA (red), Rater-1 (blue) and Rater-2 (violet).
The figures “a” to “i” are different patients.
Figure 5. Disagreement plot for the patient…
Figure 5. Disagreement plot for the patient shown in Fig. 3.
Measurements for non-enhancing T2-hyperintense tissue (NCE-T2) are indicated with circles, for contrast-enhancing tumor (CET) with crosses. Estimates of Bratumia are shown in red, of Rater-1 in blue and of Rater-2 in violet. The logarithm is used to facilitate visualisation. The zero line indicates a stable volume (i.e. Δrel = 1) from one time point to the next one. A positive value indicates a volume increase, whereas a negative value indicates a volume decrease between consecutive time points. Estimates of two raters which lie on different sides of the zero line are regarded as a disagreement between the two raters.
Figure 6. Disagreement matrix for non-enhancing T…
Figure 6. Disagreement matrix for non-enhancing T2-hyperintense tissue (NCE-T2).
Each entry reflects the total amount of disagreement between two raters. A disagreement is measured as the absolute difference between the estimates of two raters. It is measured if the estimates lie on different sides of the zero line in the disagreement plot (Fig. 5). The total amount of disagreement is the sum of disagreements over all patients.
Figure 7. Disagreement matrix for contrast-enhancing tumor…
Figure 7. Disagreement matrix for contrast-enhancing tumor (CET).
Each entry reflects the total amount of disagreement between two raters. A disagreement is measured as the absolute difference between the estimates of two raters. It is measured if the estimates lie on different sides of the zero line in the disagreement plot (Fig. 5). The total amount of disagreement is the sum of disagreements over all patients.

References

    1. Chinot O. L. et al.. Response assessment criteria for glioblastoma: practical adaptation and implementation in clinical trials of antiangiogenic therapy. Curr. Neurol. Neurosci. Rep. 13, 347 (2013).
    1. Sorensen A. G. et al.. Comparison of diameter and perimeter methods for tumor volume calculation. J. Clin. Oncol. 19, 551–557 (2001).
    1. Fraioli F. et al.. Volumetric evaluation of therapy response in patients with lung metastases. Preliminary results with a computer system (CAD) and comparison with unidimensional measurements. Radiol. Med. 111, 365–375 (2006).
    1. Marten K. et al.. Inadequacy of manual measurements compared to automated CT volumetry in assessment of treatment response of pulmonary metastases using RECIST criteria. Eur. Radiol. 16, 781–790 (2006).
    1. Kanaly C. W. et al.. A novel, reproducible, and objective method for volumetric magnetic resonance imaging assessment of enhancing glioblastoma. J. Neurosurg. 76, 1–7 (2014).
    1. Reuter M. et al.. Impact of MRI head placement on glioma response assessment. J. Neurooncol. 118, 123–129 (2014).
    1. Kanaly C. W. et al.. A novel method for volumetric MRI response assessment of enhancing brain tumors. PloS One 6, doi: 10.1371/journal.pone.0016031 (2011).
    1. Weltens C. et al.. Interobserver variations in gross tumor volume delineation of brain tumors on computed tomography and impact of magnetic resonance imaging. Radiother. Oncol. 60, 49–59 (2001).
    1. Egger J. et al.. GBM volumetry using the 3D Slicer medical image computing platform. Sci. Rep. 3, doi: 10.1038/srep01364 (2013).
    1. Menze B. H. et al.. The multimodal brain tumor image segmentation benchmark (BRATS). IEEE Trans. Med. Imaging 34, 1993–2024 (2015).
    1. Pope W. B. et al.. MR imaging correlates of survival in patients with high-grade gliomas. AJNR Am. J. Neuroradiol. 26, 2466–2474 (2005).
    1. Iliadis G. et al.. Volumetric and MGMT parameters in glioblastoma patients: Survival analysis. BMC Cancer 12, doi: 10.1186/1471-2407-12-3 (2012).
    1. Gutman D. A. et al.. MR imaging predictors of molecular profile and survival: multi-institutional study of the TCGA glioblastoma data set. Radiology 267, 560–569 (2013).
    1. Mazurowski M. A., Desjardins A. & Malof J. M. Imaging descriptors improve the predictive power of survival models for glioblastoma patients. Neuro Oncol. 15, 1389–1394 (2013).
    1. Bauer S., Nolte L.-P. & Reyes M. Fully Automatic Segmentation of Brain Tumor Images using Support Vector Machine Classification in Combination with Hierarchical Conditional Random Field Regularization. In MICCAI 2011 of LNCS Vol. 6893 (eds Fichtinger G. et al.. ), 354–361 (Springer, 2011).
    1. Zikic D. et al.. Decision forests for tissue-specific segmentation of high-grade gliomas in multi-channel MR. In MICCAI 2012 of LNCS Vol. 7512 (eds Ayache N. et al.. ), 369–376 (Springer, 2012).
    1. Meier R., Bauer S., Slotboom J., Wiest R. & Reyes M. Appearance-and context-sensitive features for brain tumor segmentation. In Proceedings of MICCAI BRATS Challenge. (2014) Available at: . (Accessed: 5th May 2015).
    1. Bauer S., Wiest R., Nolte L.-P. & Reyes M. A survey of MRI-based medical image analysis for brain tumor studies. Phys. Med. Biol. 58, 97–129 (2013).
    1. Porz N. et al.. Multi-modal glioblastoma segmentation: Man versus machine. PLoS One 9, doi: 10.1371/journal.pone.0096873 (2014).
    1. Steed X. T. C. et al.. Iterative probabilistic voxel labeling : Automated segmentation for analysis of the cancer imaging archive glioblastoma Images. AJNR Am. J. Neuroradiol. 36, 678–685 (2015).
    1. Weizman L. et al.. Semiautomatic segmentation and follow-up of multicomponent low-grade tumors in longitudinal brain MRI studies. Med. Phys. 41, doi: 10.1118/1.4871040 (2014).
    1. Liberman G. et al.. Automatic multi-modal MR tissue classification for the assessment of response to bevacizumab in patients with glioblastoma. Eur. J. Radiol. 82, 87–94 (2013).
    1. MacDonald D., Cascino T., Schold S. J. & Cairncross J. Response criteria for phase II studies of supratentorial malignant glioma. J. Clin. Oncol. 8, 1277–1280 (1990).
    1. Bauer S., Fejes T. & Reyes M. A skull-stripping filter for ITK. Insight J. 20, 1–7 (2012).
    1. Jakab A. Segmenting brain tumors with the slicer 3D software. Technical Report. (2012) Available at: . (Accessed: 11th May 2015).
    1. Fedorov A. et al.. 3D Slicer as an image computing platform for the quantitative imaging network. Magn. Reson. Imaging 30, 1323–1341 (2012).
    1. Wen P. Y. et al.. Updated response assessment criteria for high-grade gliomas: Response assessment in neuro-oncology working group. J. Clin. Oncol. 28, 1963–1972 (2010).
    1. Tustison N. J. et al.. N4ITK: Improved N3 bias correction. IEEE Trans. Med. Imaging 29, 1310–1320 (2010).
    1. Nyul L. G., Udupa J. K. & Zhang X. New variants of a method of MRI scale standardization. IEEE Trans. Med. Imaging 19, 143–150 (2000).
    1. Breiman L. Random forests. Mach. Learn. 45, 5–32 (2001).
    1. Criminisi A. & Shotton J. In Decision Forests for Computer Vision and Medical Image Analysis, (eds Criminisi A. et al.. ), 25–45 (Springer, 2013).
    1. Mazzara G. P., Velthuizen R. P., Pearlman J. L., Greenberg H. M. & Wagner H. Brain tumor target volume determination for radiation treatment planning through automated MRI segmentation. Int. J. Radiat. Oncol. Biol. Phys. 59, 300–312 (2004).
    1. Galanis E. et al.. Validation of neuroradiologic response assessment in gliomas: measurement by RECIST, two-dimensional, computer-assisted tumor area, and computer-assisted tumor volume methods. Neuro Oncol. 8, 156–165 (2006).
    1. Shah G. D. et al.. Comparison of linear and volumetric criteria in assessing tumor response in adult high-grade gliomas. Neuro Oncol. 8, 38–46 (2006).
    1. Pirzkall A. et al.. Tumor regrowth between surgery and initiation of adjuvant therapy in patients with newly diagnosed glioblastoma. Neuro Oncol. 11, 842–852 (2009).
    1. Ellingson B. M. et al.. Quantitative volumetric analysis of conventional MRI response in recurrent glioblastoma treated with bevacizumab. Neuro Oncol. 13, 401–409 (2011).
    1. Stummer W. et al.. Extent of resection and survival in glioblastoma multiforme: Identification of and adjustment for bias. Neurosurgery 62, 564–574 (2008).
    1. Jakola A. S. et al.. Surgically acquired deficits and diffusion weighted MRI changes after glioma resection - A matched case-control study with blinded neuroradiological assessment. PLoS One 9, doi: 10.1371/journal.pone.0101805 (2014).
    1. Grabowski M. M. et al.. Residual tumor volume versus extent of resection: predictors of survival after surgery for glioblastoma. J. Neurosurg. 121, 1115–1123 (2014).
    1. Zinn P. O. et al.. Radiogenomic mapping of edema/cellular invasion MRI-phenotypes in glioblastoma multiforme. PLoS One 6, doi: 10.1371/journal.pone.0025451 (2011).
    1. Naeini K. M. et al.. Identifying the mesenchymal molecular subtype of glioblastoma using quantitative volumetric analysis of anatomic magnetic resonance images. Neuro Oncol. 15, 626–634 (2013).
    1. Rios Velazquez E. et al.. Fully automatic GBM segmentation in the TCGA-GBM dataset : Prognosis and correlation with VASARI features. Sci. Rep. 5, doi: 10.1038/srep16822 (2015).
    1. Cordova J. S. et al.. Quantitative tumor segmentation for evaluation of extent of glioblastoma resection to facilitate multisite clinical trials. Transl. Oncol. 7, 40–47 (2014).
    1. Meier R., Bauer S., Slotboom J., Wiest R. & Reyes M. Patient-specific semi-supervised learning for postoperative brain tumor segmentation. In MICCAI 2014 of LNCS Vol. 8673 (eds Golland P. et al.. ), 714–721 (Springer, 2014).
    1. Bauer S., Tessier J., Krieter O., Nolte L. P. & Reyes M. Integrated spatio-temporal segmentation of longitudinal brain tumor imaging studies. In MICCAI 2013 MCV Workshop of LNCS Vol. 8331 (eds Menze B. et al.. ), 74–83 (Springer, 2014).
    1. Altman D. G. & Bland J. M. Measurement in medicine : the analysis of method comparison studies. Statistician 32, 307–317 (1983).
    1. Bauknecht H.-C. et al.. Intra- and interobserver variability of linear and volumetric measurements of brain metastases using contrast-enhanced magnetic resonance imaging. Invest. Radiol. 45, 49–56 (2010).
    1. Mantel N. The detection of disease clustering and a generalized regression approach. Cancer Res. 27, 209–220 (1967).

Source: PubMed

Sponsorzy i współpracownicy

Warunki medyczne

Interwencje lekowe

3
Subskrybuj